![scusa.jpg](https://capedge.com/proxy/CORRESP/0001580608-17-000088/scusa.jpg)
VIA EDGAR
October 27, 2017
Ms. Stephanie L. Sullivan
United States Securities and Exchange Commission
Division of Corporation Finance
100 F Street, N.E.
Washington, D.C. 20549
Re: Santander Consumer USA Holdings Inc.
Form 10-K for the Fiscal Year Ended December 31, 2016
Filed February 28, 2017
Form 10-Q for the Fiscal Quarter Ended September 30, 2017
Filed August 2, 2017
File No. 001-36270
Dear Ms. Sullivan:
This letter includes the responses of Santander Consumer USA Holdings Inc. (the “Company”, “SC” or “Management”) to the comments of the staff of the Securities and Exchange Commission (the “Commission”), Division of Corporation Finance (the “Staff”), regarding its Form 10-K for the fiscal year ended December 31, 2016 (the “Form 10-K”) and its Form 10-Q for the quarter ended June 30, 2017 (the “Form 10-Q”), which were delivered in a letter dated September 28, 2017. For your convenience, in this letter the text of the Staff’s comments is set forth in bold, italicized text followed by the responses of the Company.
Form 10-K for the Fiscal Year Ended December 31, 2016
Item 7. Management’s Discussion and Analysis of Financial Condition and Results of Operations, page 46
Year Ended December 31, 2015 Compared to Year Ended December 31, 2014, page 60
Provision for Credit Losses, page 61
1. | We note that on page 59 of the 2015 Form 10-K/A, you disclose that your credit loss forecasting models were not calibrated for a higher concentration of loans with limited bureau information and that you recorded a $149 million qualitative adjustment to accommodate additional expected charge offs. We also note that on page 61 of your 2016 Form 10-K, you make reference to this qualitative adjustment recorded during 2015, but do not indicate whether such a qualitative adjustment continues to be recorded. Please respond to the following: |
• | Tell us whether the models have been re-calibrated in 2016 or 2017. If so, tell us when and describe for us the modeling and other changes made to capture this in your credit loss forecasting models. |
Ms. Stephanie L. Sullivan
October 27, 2017
Page 2
Response:
We included a qualitative adjustment of $149 million in the allowance for credit losses as of December 31, 2015 to address changes in our underwriting practices since early 2015 that were likely to cause estimated credit losses to differ from historical loss experience. Specifically, the qualitative adjustment reflected a higher concentration of loans with limited bureau information (“thin files1”) in the portfolio than was being captured in the model historical loss data. As additional historical experience was incorporated into our vintage loss model through 2016, the qualitative adjustment for thin files dissipated over 2016 and was reduced to zero as of September 30, 2016.
We implemented a new model (“Markov”) as of June 30, 2017 that replaced our vintage loss model (further explained in response to Question 13 below). The Markov model uses transition matrices that are categorized based on account characteristics. One of the account characteristics is a “Thin File Indicator”, which ensures that we have appropriately captured the losses from thin files in our total credit loss forecasting model.
• | Tell us whether you took a qualitative adjustment in either 2016 or 2017 related to this matter. If so, please quantify for us the amount of the adjustments, and describe the changes in the level of the adjustment between 2015 and 2017, as applicable. |
Response:
As noted above, “thin files” is an account characteristic that is used in our model to estimate the allowance for credit losses. No qualitative or other unique model adjustments have been warranted other than discussed above. The Company did not include any qualitative reserves for this matter as of December 31, 2016, March 31, 2017 and June 30, 2017.
Credit Quality, page 64
Deferrals and Troubled Debt Restructurings, page 66
2. | We note your disclosure that at the time a deferral is granted, all delinquent amounts may be deferred or paid, resulting in the classification of the loan as current and therefore not considered a delinquent account. Please respond to the following: |
• | Tell us whether there are instances where you would not consider all delinquent amounts current at the time the deferral is granted, and quantify those scenarios relative to the scenarios where all delinquent amounts are considered current. |
Response:
In general, loans are brought “current” upon the first deferral. However, there are instances when a deferment is granted but the loan is not brought completely current. For example, if a borrower is 65 days past due (“DPD”) at the time deferral is approved, and a two month deferment2. is processed (~60 days), the deferral will keep the account 5 days past due.
1Represents loans with less than 4 trade lines or less than 36 months of credit bureau history
2Over 90% of deferrals granted are for only two months
Ms. Stephanie L. Sullivan
October 27, 2017
Page 3
Historically, a majority of deferrals are done for borrowers in the 31-60 and 61-90 past due buckets. Based on the number of days deferred, approximately 80% of the borrowers are reported as “current” after deferral and approximately 20% of borrowers remain delinquent. Further, a customer is limited to one deferral each 6 months, and if a customer receives a certain number3 of deferrals over the life of the loan, the loan will advance to a TDR designation.
• | Tell us whether you require a period of performance under the deferral or modified terms prior to deciding whether to consider all delinquent amounts current. |
Response:
A period of performance under the deferral or modified terms is not used as the determinative factor to classifying the amount as current. Rather, we consider a number of factors regarding the borrower and payment history in determining whether the borrower is expected to make the required payments and a deferral should be granted.
Before granting a deferral or modified terms, we use a proprietary model that evaluates the consumer’s credit profile, payment history, and capacity to perform under the extended payment terms. Our analysis supports our expectation of future collections on these loans and supports reporting the account in current status.
In addition, periodically, we evaluate historical performance and have noted that 75-80% of loans that receive deferrals or other modifications in any month, continue to make interest and principal payments and therefore accrue interest six months following the deferral.
Note, a deferral does not prevent either a non-accrual or a TDR designation, in that a certain number of deferrals will require a loan to be placed on non-accrual and / or otherwise classified as a TDR.
3. | We note that the level of deferrals on retail installment contracts that have been deferred twice, deferred 3-4 times, and deferred greater than 4 times continues to increase each period, such that at June 30, 2017, these deferrals make up 7.7%, 8.6% and 0.3% of your deferral population. Please respond to the following: |
• | In light of the significance of the deferrals which are considered troubled debt restructurings (TDRs), and the subsequent high levels of those which require further deferral after that, please tell us why you only consider a TDR to have subsequently defaulted upon charge-off at the earlier of the date of repossession or 120 days past due (per page 108). |
Response:
As noted in our response to question 2 above, deferrals are granted to borrowers only under specific guidelines and, as a result, a majority of loans that have received a deferral (many of which are also already designated as TDRs) continue to make consistent payments. However,
3The Company considers TDRs to include all individually acquired retail installment contracts or personal revolving loans that have been modified at least once, deferred for a period of 90 days or more, or deferred at least twice.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 4
given our repossession practices a significant number of at risk loans actually charge-off prior to 120 DPD due to default and the resulting repossession. TDRs that subsequently default are appropriately considered in our credit loss allowance methodology for the TDR loan portfolio. Refer to response to next question.
• | We note your disclosure on page 24 of your June 30, 2017 Form 10-Q that for TDR modifications, and for TDR modifications that subsequently default, the allowance methodology remains unchanged. Please tell us how you factor in the likelihood of default in your TDR allowance methodology, including the data considered in this analysis, in light of the high level of subsequent deferrals which would not be considered defaults of the original TDR based on your definition. |
Response:
Our credit loss allowance methodology for the TDR loan portfolio is based on the present value of expected cash flows under ASC 310-10 and ASC 310-40. TDR loans that have common risk characteristics are pooled to project future cash flows, as permitted under ASC 310-10-35-21. Cash flows are estimated based on the historical performance and loan transition from performing to delinquency statuses and ultimately to loss, across 19 categories. The model considers loan delinquency status, TDR type (deferral, modification, etc.), internal credit risk score, thin/thick file and time since TDR event. For loans captured in each of these categories the estimated allowance appropriately varies. For example, based on Q2’2017 results, the ALLL coverage ratio4 across the 19 categories varied between 13% and 40%.
Note 1 - Description of Business, Basis of Presentation, and Significant Accounting Policies and Practices, page 89
Retail Installment Contracts, page 90
4. | We note your disclosure that you estimate future principal prepayments in the calculation of the constant effective yield. Please respond to the following: |
• | Tell us the levels used for purposes of estimating the constant effective yield. For example, describe whether you aggregate loans by period of origination, collateral type or some other fashion. |
Response:
We pool loans into multiple categories for the purposes of anticipating prepayments and estimating the constant effective yield in accordance with ASC 310-20-35-30: first, loans are pooled by monthly vintage (period of origination); then within the vintage, loans are further pooled by both term (contractual maturity) and Loss Forecasting Score5. (LFS, which is used to rank order the credit quality of the loans at origination.) Prepayments in our portfolio are sensitive to credit
4As of June 30, 2017, the average TDR Allowance ratio was 28.7% (i.e. TDR impairment as a % of TDR unpaid principal balance)
5LFS is an SC proprietary calculation utilized by Company to assess the credit quality of a particular loan. It is driven by relevant credit metrics (i.e. FICO, LTV, etc.). Lower LFS scores generally represent lower credit quality loans.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 5
quality, with higher credit quality loans generally experiencing higher voluntary prepayment rates
than lower credit quality loans. The impact of defaults is not considered in the prepayment rate; the prepayment rate only considers voluntary prepayments. As of June 30, 2017, unique effective yields were computed for approximately 500 pools of loans, with three credit quality bands, two maturity bands and approximately 80 monthly vintages that each have a unique prepayment curve.
Given the large number of SC loans with substantially similar credit characteristics and behaviors, we consider our pools homogenous and maintain that prepayment activity is reasonably predictable for each pool.
We have consistently utilized historical data to estimate the timing and the amounts of prepayments, and our model adjusts for differences between prior forecasts and actuals on a monthly basis.
• | Please tell us, and revise future filings to disclose, the quantitative assumptions used in estimating the prepayment estimate, in accordance with ASC 310-20-50-2. |
Response:
We intend to enhance disclosure in future filings as set forth below (new language underlined):
The amortization of discounts, subvention payments from manufacturers, and other origination costs on retail installment contracts held for investment acquired individually, or through a direct lending program, are recognized as adjustments to the yield of the related contract using the effective interest method. The Company estimates future principal prepayments specific to pools of homogenous loans which are based on the vintage, credit quality at origination and term of the loan. The resulting prepayment rate specific to each pool is, based on historical experience, and is used as an input in the calculation of the constant effective yield.
Leased Vehicles, net, page 92
5. | We note your disclosure that the expected residual value, and accordingly the monthly depreciation expense may change throughout the term of the lease. Your disclosure goes on to state that you estimate expected residual values using independent data sources and internal models that take into consideration economic conditions, current auction results, your remarketing abilities and manufacturer vehicle and marketing programs. Given your disclosures regarding the ongoing decline in used vehicle sale prices, as discussed on pages 7-10, please tell us whether you have made any changes to your depreciation rates since January 1, 2015. If so, please quantify those changes, the class of leased assets impacted, and clearly indicate the periods the changes were effective. |
Response:
We depreciate the investment in operating leases over the contractual term of the lease to the expected residual value.
The residual valuation testing that we apply uses two sources, our proprietary modeled forecast and the Automotive Lease Guide “ALG”6 estimate, applied at an individual asset level to provide an
6ALG , Inc. is a third-party company providing Residual Value projections to the automotive industry
Ms. Stephanie L. Sullivan
October 27, 2017
Page 6
estimate of residual value. During the residual value validation process, the lower valuation between the two sources is typically the residual value utilized.
At lease end, vehicles with market rates higher than contracted residual are often purchased by the dealer or customer, while vehicles with market rates lower than contracted residual are generally returned to SC. This consumer tendency means SC has a higher chance to receive more of the vehicles with stressed values than non-stressed, making the use of the lower of the two sources most appropriate.
We re-estimate our residual values on a monthly basis, and use the results to adjust depreciation expense on a lease level basis. Lease vehicle expense, as reported in our SEC filings, includes (a) net depreciation expense; (b) gain or loss on sale of lease vehicle; and (c) net charge off balance.
Net depreciation represents (a) depreciation based on contractual residual values; and (b) increase/decrease in depreciation based on changes to residual values, as reduced by the amortization of the subvention incentives7 received from the manufacturer.
Changes to residual values driven by lower used car prices have contributed to an increase in net depreciation expense recorded, compared to the originally estimated depreciation, of 6% in 2015, 4% in 2016, and 9% June 2017 year-to-date.
6. | We note your disclosure that you periodically evaluate your investment in operating leases for impairment if circumstances, such as a general decline in used vehicle values, indicate that impairment may exist. However, we also note your disclosure on page 79 that impairment of an operating lease asset is assessed upon the occurrence of a triggering event, which you disclose are systemic, observed events impacting the used vehicle market, such as shocks to oil and gas prices. Lastly, you state that you have not taken any such impairment charges. Please respond to the following: |
• | In light of the similar reasons for adjusting depreciation expense, and for assessing the operating lease for impairment, please tell us in more detail how you would evaluate whether to adjust future depreciation rates versus take an impairment charge. |
Response:
See response to comment 5 above, for our methodology on adjusting depreciation on a monthly basis.
In addition to accelerating depreciation on all of our lease portfolio, we perform an impairment assessment for our investment in operating leases, on a quarterly basis, by assessing whether the sum of the undiscounted expected future cash flows is less than the carrying amount of the investment.
• | Given your disclosure of the ongoing decline in used vehicle sales prices on pages 7-10, please tell us whether you have performed an impairment assessment for your investment in operating leases. |
7We receive manufacturer incentives used as a component difference between the projected ALG residual value and the contractual residual value
Ms. Stephanie L. Sullivan
October 27, 2017
Page 7
Response:
We perform an impairment assessment for our investment in operating leases for impairment, on a quarterly basis, consistent with requirements of ASC 360-10- 35-21. Based on our analysis at December 31, 2016 and June 30, 2017, we concluded that there were no impairment indicators, which might require a recoverability analysis of leased vehicles. Refer to response to next question regarding how we consider changes in residual values.
• | Tell us whether you have a specific threshold that is considered when evaluating the size of the general decline in used vehicle values and the necessity for performing an impairment assessment. |
Response:
We regularly monitor changes in residual value forecasts at a vehicle model level. When model level projected average residual values drop significantly, defined as a greater than 20% decline (absent any accelerated depreciation as discussed above) quarter-over-quarter8, a recoverability analysis for the applicable model(s) is performed.
We established this threshold as a result of a recent internal study that assessed the lease portfolio’s sensitivity to declines in residual values. The study considered twelve years of actual used car market prices vs. ALG projections and applied it to our portfolio vehicle mix. The study concluded that the portfolio would have to experience a 21.8% decline in residual values sold at Physical Auction, higher than ever historically reported, for an impairment to occur. Key assumptions in the analysis are no mitigants (such customers that terminate their leases early are responsible for any residual losses) to residual value loss as well as FCA9 being unable to meet its residual risk share guarantees10. Assuming FCA fulfilled its residual risk share guarantees, a residual shock in excess of 30% would be required for an impairment to occur.
Note 9 - Other Assets, page 119
7. | We note that the upfront fee related to the Chrysler agreement totaled $95.0 million as of December 31, 2016. We also note the significant disparity between the penetration standards agreed to with Chrysler, and the actual penetration rates, which, other than program year 1, have always remained well below the target penetration rate, as disclosed on page 46. Please respond to the following: |
• | Tell us whether you performed an impairment assessment for this upfront fee in light of the fact that you have been unable to meet the penetration targets, and the fact that the disparity between the targets and actual have only continued to grow. If so, please tell us how this assessment was performed, and if not, please tell us why. |
8Since we accelerate our depreciation each month based on revised residual values at lease level, the risk of impairment is minimized.
9Fiat Chrysler Automobiles US LLC, formerly Chrysler Group LLC
10The Company and Chrysler bear Residual Value Losses, defined as the excess of originally estimated residual value (contract residual less residual enhancement subvention, if any) over actual gross sales proceeds from disposition of leased vehicles.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 8
Response:
We paid a $150 million upfront fee at the inception of the Chrysler agreement in May 2013. The fee is being amortized into income over a ten year term. As of December 31, 2016 and June 30, 2017, the unamortized balance of the upfront fee related to the Chrysler agreement totaled $95 million and $87.5 million, respectively. While we do not view this arrangement as a financial reporting segment, the performance from actual origination volume and related servicing costs has been accretable to net income, and in excess of the minimum return on investment that we expected.
We assess whether impairment indicators are present for the carrying value of the upfront fee related to the Chrysler agreement, in accordance with ASC 360 Property, Plant and Equipment11, by evaluating the following factors:
a) | adverse change in the extent or manner in which the prepaid asset is being used; |
b) | inability to meet penetration targets; |
c) | current-period operating or cash flow loss; |
d) | accumulation of associated costs with the expected benefits of the Chrysler relationship that would render the prepaid unrecoverable; |
e) | expected sale or disposal of the asset. |
Of the considerations above, the only factor that has not performed at, or above, the original expectations is our inability to meet the penetration targets. The program has been profitable for us and for Chrysler and there are no contractual lump sum, future origination loan or subvention pricing concessions, or changes in other contractual terms that are punitive to either party in consideration of the planned penetration rates.
Therefore, there are no triggering events to indicate that the carrying should not continue to be expensed ratably over the term of the agreement and therefore the upfront fee need not be tested for recoverability as of December 31, 2016 and June 30, 2017.
• | We note your disclosure on page 46 that Chrysler Capital continues to be focal point of your strategy, and that you remain committed to the Chrysler agreement. However, we also note your disclosure on page 123 that Chrysler may terminate the agreement if you do not meet the penetration rates for the first five years of the agreement. Tell us whether Chrysler would be able to keep the existing upfront fee if they chose to terminate the contract due to the targets not being met. |
Response:
The upfront fee paid to Chrysler is a nonrefundable fee from a contractual standpoint and SC would not be entitled to receive the fee back if Chrysler decides to terminate the agreement. Chrysler can advise us that they intend to terminate due to penetration rates, but we believe that Chrysler must treat us in a manner consistent with how other manufacturers treat their captives in order for the penetration rates clause to be enforceable.
11The guidance in the Impairment or Disposal of Long-Lived Assets under ASC 360-10-15-4, includes long-term prepaid assets.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 9
• | Tell us whether Chrysler has provided you with any indications that it is considering canceling or amending this agreement. |
Response:
We don’t believe that Chrysler plans to terminate the agreement with us. In fact, we continue to work constructively with Chrysler in order to identify common strategic opportunities for growth and have taken deliberate steps together collaboratively, to attempt to enhance our penetration rates, to the benefit of both parties.
Note 11 - Commitments and Contingencies, page 123
Legal and Regulatory Proceedings, page 124
8. | We note your disclosure that you are unable to reasonably estimate your potential exposure, if any, to the lawsuits, regulatory matters and other legal proceedings at this time, but it is reasonably possible that actual outcomes or losses may differ materially from your assessment, and adverse resolution of any of the matters could have a material adverse effect on your financial position, liquidity, and results of operation. However, we also note that in prior filings, including your September 30, 2016 Form 10-Q, and your December 31, 2015 Form 10-K/A, you conclude that you do not believe there are any proceedings, threatened or pending, that if determined adversely, would have a material adverse effect on your consolidated financial position, results of operations, or liquidity. Please respond to the following: |
• | Given the change in your disclosure where you now conclude that adverse resolution of any of the matters could have a material effect to you, please tell us how you concluded that you cannot estimate the reasonably possible additional loss or range of loss, for at least certain of the matters. |
Response:
In late 2014, we received subpoenas and/or civil investigative demands, from the Attorneys General of California, Illinois, Oregon, New Jersey, Maryland and Washington12 under the authority of each state's consumer protection statutes as disclosed in our Form 10-K and Form 10-Q. These six states’ Attorneys General serve as an executive committee on behalf of a group of 29 states (the “Consortium”). The Consortium is led by the Illinois Attorney General. The information requests were broad and included, among other things, request for documents pertaining to the Company’s underwriting, dealer management, loan acquisition and securitization activities. We have responded to all requests.
The Plaintiff/Consortium has provided an overview of its analysis of the documents produced and the testimony of the Company’s officers, and have also made a non-specific allegation of liability, generally focused on the high default rates in the Company’s portfolio. However, the Plaintiff /Consortium has not defined a class of consumers that it believes has been harmed by the Company and has not specified any actions by the Company which may violate a law.
12These states will serve as an executive committee on behalf of a group of 30 state Attorneys General (collectively the “Consortium”)
Ms. Stephanie L. Sullivan
October 27, 2017
Page 10
Materiality
We have determined that the investigation (the “Matter”) is a material event. Our practice is for any matter for which we receive notice of a claim, we will make a determination about whether the matter, if unfavorably resolved, could expose us to a risk of loss.
Probability of loss
We believe that it is reasonably possible that we may suffer a loss related to the Matter. We define “Reasonably Possible” as a loss that falls in between the range of “Remote” and “Probable”. While the Consortium has not articulated a class of customers impacted by our purported conduct or asserted that our actions violated a specific law, based on our experience with other regulatory matters, we believe it is reasonably possible we will suffer a loss.
Estimate of loss
We do not believe that any loss we could incur is reasonably estimable. Since the Consortium has not defined a class of consumers it believes has been harmed and has not specified actions by SC which may violate the law, we are unable to estimate a loss for the Matter.
The potential claims by the Consortium are a product of individual states’ consumer protection laws and, as such, the theory of liability and any potential liability associated with purported violations of those laws is not necessarily the same. In addition, the identification of the class of consumers allegedly impacted by SC’s purported conduct, once articulated, may be different from the group of consumers included, even when compared to the previously disclosed settlement with the Massachusetts AG.
Due to these factors, we disclosed that it is reasonably possible the settlement of these matters may result in a loss but that the loss or range of loss is not currently estimable.
• | If you continue to believe that you cannot estimate the reasonably possible additional loss or range of loss, please supplementally: (1) explain to us the procedures you undertake on a quarterly basis to attempt to develop a range of reasonably possible loss for disclosure and (2) for each material matter, what specific factors are causing the inability to estimate and when you expect those factors to be alleviated. We recognize that there are a number of uncertainties and potential outcomes associated with loss contingencies. Nonetheless, an effort should be made to develop estimates for purposes of disclosure, including determining which of the potential outcomes are reasonably possible and what the reasonably possible range of losses would be for those reasonably possible outcomes. |
Response:
On a periodic basis, we obtain a legal opinion from outside counsel regarding the legal matters that meet our materiality thresholds. We also conduct our own assessment of matters that meet the materiality threshold to determine how the matters should be disclosed in our Form 10-K and 10-Q.
The Company notes that there is only one matter (discussed above) that is driving our disclosure. In relation to the Matter discussed above, we will continue to evaluate whether the potential impact of the Matter can be estimated. We believe that a loss estimate may be possible when the Consortium defines a specific class of customers that has been harmed and articulates conduct of
Ms. Stephanie L. Sullivan
October 27, 2017
Page 11
SC which the Consortium alleges violates a specific law; or if another suit is filed with respect to the general allegations by the Consortium or an individual state.
Note 18 - Investment Gains/Losses
9. | We note that lower of cost or market adjustments on your personal loans are principally driven by customer default activity. Specifically, you recognized $429.1 million of lower of cost or market adjustments due to customer default activity, out of $423.6 million of total lower of cost or market adjustments during the year ended December 31, 2016. We also note your disclosure on page 137 that these personal loans held for sale are valued based on significant unobservable inputs on key assumptions, including historical default rates and credit loss expectations. Please respond to the following: |
• | Tell us whether you are estimating customer default activity, or whether you are measuring the impact of the defaults as they occur. |
Response:
As of December 31, 2016 and as of June 30, 2017 our personal lending portfolio primarily pertains to Bluestem Brands, Inc. (“Bluestem”), an online retailer for whose customers we provide financing mainly in the form of credit card/revolving line. We classify the Bluestem portfolio as held for sale and report at lower of cost or fair value (“LOCOM”) in accordance with ASC 310-35-48.
The valuation of the Bluestem portfolio is based on a combination of market approach and income approach, which takes into consideration net credit loss expectations estimated using historical customer default activity, as well as market participant pricing factors.
The Investment gains / losses balance represents (a) the change in the LOCOM based on the estimation of future customer default activity; and (b) difference between actual and estimated losses during the period.
• | In light of the fact that your gross personal loan balances continues to decrease, please describe for us in further detail the significant drivers of the continued write-downs due to customer default activity occurring during 2016 and the six months ended June 30, 2017. |
Response:
The key driver to continued write-downs is the LOCOM adjustment recorded for each new originated loan, based on forecasted lifetime loss.
• | Identify the specific assumptions that have the most significant impact on the write-down to fair value and tell us whether they are one of the unobservable inputs to the fair value disclosed on page 49 of your June 30, 2017 Form 10-K (and which were not disclosed in the 2016 Form 10-K). If so, tell us the comparable inputs to the fair value during 2015 and 2016. |
Response:
Ms. Stephanie L. Sullivan
October 27, 2017
Page 12
The specific assumptions that have the most significant impact on the write-down to fair value are as follows:
a) | Principal cash flows and net payment rate: Monthly principal cash flows were projected based on the net payment rate until the current outstanding principal balance is fully paid off. The net payment rate was applied to the beginning principal outstanding balance and was based on historical and forecasted net payment rates of the loans. |
b) | Net credit adjustment: The cash flows were adjusted based on our estimated cumulative net credit loss expectations, which were determined by historical customer default activity. |
c) | Discount rate: The discount rate was developed based on the relative risk of cash flows, taking into consideration market participant expectations and inclusive of the loan type, market rates as of valuation date, credit expectations and capital structure. |
d) | Market participant view: The actual bids received from various market participants. |
We acknowledge that our disclosures do not include “credit default assumptions” as one of the most significant unobservable inputs. As we work to continuously improve our disclosures, we intend to enhance disclosure in future filings to include “Market Approach/Income approach13” as the valuation technique and update the ranges for discount rates and credit default assumptions.
The draft disclosure to be included in our Form 10-Q for the nine months ended September 30, 2017 is:
Financial Instrument | Fair Value at September 30, 2017 (Amount in thousands) | Valuation Technique | Unobservable Inputs | Range |
Personal loans held for sale | $929,549 | Market / Income Approach | Discount Rate Default Rate | 15% - 30% 30% - 40% |
The comparable inputs to the fair value during 2015 and 2016, including the range for discount rates and credit default assumptions, were consistent with the draft disclosure above.
Form 10-Q for the Fiscal Quarter Ended June 30, 2017
Note 4 - Credit Loss Allowance and Credit Quality, page 19
10. | We note your disclosure on page 21 regarding the differentiation between required minimum payment amounts for receivables originated before and after January 1, 2017 for purposes of determining delinquency statistics. Specifically, you revised the required minimum payment to be 90% of the scheduled payment from the previously used 50% of the scheduled payment for receivables acquired by the Company from an unaffiliated third party originator or originated by the Company. |
13 As compared to “Market Approach” disclosed in our Q2’17
Ms. Stephanie L. Sullivan
October 27, 2017
Page 13
• | Tell us the reason for this change and how you considered whether it constituted a change in accounting policy or principle or a correction of an error under ASC 250. |
Response:
During 2016, we began to note some deterioration in the performance of recent originations, particularly those loans originated in 2015, and addressed those trends with the introduction of more disciplined underwriting standards in late 2016. In January 2017, we had more complete information on the deteriorating performance and concurrent with those assessment we also chose to update our servicing practices regarding the required minimum payment from 50% to 90% for our Core14 portfolio.
All contracts for loans originated after January 1, 2017 were updated to include these new minimum payment requirements and, accordingly, our accounting results and associated disclosures were updated.
We evaluated the nature of the contracts and related servicing practices for customer loans originated after January 1, 2017 and determined that this was neither (a) a change in policy or principle; nor (b) a correction of an error, under ASC 250. This is the application of our existing accounting policies and principles to account for newly originated loans with servicing practices that are distinctly different from loans originated prior to January 1, 2017. This change was driven by a change in the underlying facts and circumstances specific to borrower cash flows:
1) | Lending to Customers: Our Core customers are aware of our servicing practices relative to delinquency, credit bureau reporting and deferral processing. We systemically adjusted our pre-2017 originations and our post-2017 originations to easily communicate with our customers regarding payment thresholds. We also have multiple methods of communications (mail, fax, email, and website) for our customers to see and assist in explaining the effect of deferments and partial payments have on their accounts. |
2) | Funding our Lending to Customers: Our securitization structures are specific as to the servicing of the loans and measurement of delinquencies, among other specific applications of cash received from customers. In addition to disclosing this change in our Q1’2017 financial statement, these changes have been incorporated in our 2017 securitizations as well. For example, see an extract from the Offering memorandum supplement for DRIVE 2017-A: |
“With respect to receivables originated by SC prior to January 1, 2017 and through its “Chrysler Capital” channel, the required minimum payment is 90% of the scheduled payment. With respect to all other receivables originated by SC or acquired by SC from an unaffiliated third-party originator prior to January 1, 2017, the required minimum payment is 50% of the scheduled payment. With respect to receivables originated by SC or acquired by SC from an unaffiliated third-party originator on or after January 1, 2017, the required minimum payment is 90% of the scheduled payment, regardless of which channel the receivable was originated through.”
14Non-Chrysler Capital retail installment contract
Ms. Stephanie L. Sullivan
October 27, 2017
Page 14
3) | This change now aligns all of our servicing practices across Core, Chrysler, and other loans serviced for others. It also aligns the application of customer cash, irrespective of the lending channel, with respect to our securitization platforms. |
4) | Lastly, notwithstanding the understanding we have with our customers about the application of cash to their loan account, aligning this particular servicing practice across our portfolio benefits our parent company since this practice is more commonly applied by regulated entities and by entities regulating consumer lending practices. |
• | Tell us whether you performed an analysis to evaluate the significance of this change to your delinquency statistics for prior periods. If so, please tell us the results of this analysis. |
Response:
We did not specifically recast any prior periods since the application of payments affects borrower behaviors and our servicing with respect to granting deferrals or other modifications.
To further clarify the relative impact to a customer account, for Core portfolio loans originated before January 1, 2017, a partial payment between 50% and 90% of the scheduled payment is counted as a whole payment, meaning the account will not roll into past due status. However, the payment following the partial payment must be a full payment, or the account will move into delinquency status at that time.
We would generally expect slightly higher delinquency in the 1-30 bucket, but would not expect higher delinquencies in later delinquency buckets. We monitor the delinquency status regularly and have not observed any unusual variations or other matters which would cause us to believe prior period customer delinquencies are not demonstrative of actual customer behavior.
• | In light of the fact that delinquencies are a direct input to your allowance methodology, as disclosed on page 91 of your Form 10-K, please tell us whether you evaluated the impact of this change to your allowance methodology. |
Response:
We reviewed the impact the servicing change may have on the amount and timing of loss recognition. Our review included reviewing the volume of accounts making payments in the 50-90% range each month, their delinquency status month over month, and also comparing like accounts between the Chrysler and Core populations. Based on our analysis of loans with similar credit quality and attributes, the gross loss rates in the first 12 months of each vintage for the loans that followed the 50% minimum payment policy was approximately 8.96% as compared to 9.01% for the loans that followed the 90% minimum payment policy. Accordingly, we concluded that the impact of this change was not significant to our total allowance.
11. | We note your disclosure on page 21 and elsewhere in the filing that retail installment contracts and personal amortizing term loans are classified as non-performing when they are greater than 60 days past due as to contractual principal or interest payments. You also state on page 58 that you generally discontinue and reverse accrued income once a loan becomes more than 60 days past due, with certain exceptions. However, footnotes to your tabular delinquency disclosures, such as on page 21, state that interest is accrued until 60 days past due, and delinquencies in your filing are |
Ms. Stephanie L. Sullivan
October 27, 2017
Page 15
shown for balances over 59 days past due. This seems inconsistent with your narrative, and the quantitative presentation appears to differ from your approach in the 2016 10-K, in which you show delinquencies in buckets of 0-60 days past due and over 60 days past due, versus 31-59 days past due and over 59 days past due in the Form 10-Q. In addition, the year-end delinquency balances shown in the June 30, 2017 Form 10-Q appear to differ from those shown on page 105 of the 2016 Form 10-K. Please address the following:
• | Tell us why the total delinquent principal amounts shown as of December 31, 2016 differ from what was disclosed in the 2016 Form 10-K. For example, total delinquent principal was $4.1 billion as of December 31, 2016 in the Form 10-K, but $4.5 billion in the Form 10-Q. |
Response:
We adjusted our delinquency disclosures effective January 1, 2017, to align with the SEC and regulatory reporting disclosures of our majority shareholder (Santander Holdings USA, Inc. or “SHUSA”). SHUSA uses the following days’ delinquent buckets for delinquency reporting for both SEC and the Federal Reserve: 0-29, 30-59, 60-89, 90+, however we historically reported the following delinquency buckets: 0-30, 31-60, 61-90, 91+.
The disclosure difference in current delinquencies buckets and earlier delinquencies buckets is only related to a one day shift.
• | Tell us whether the differences between the December 31, 2016 delinquency amounts shown for each category in the Form 10-K and the Form 10-Q are driven by the one day change in delinquency buckets. |
Response:
Yes, the difference between the December 31, 2016 delinquency amounts shown for each category in the Form 10-K and the Form 10-Q are driven by the one day change in delinquency buckets.
• | Explain the driver for the change in the tabular bucket breakout. |
Response:
Refer above.
• | Clarify your nonaccrual policy and what happens to loans that are 60 days past due. |
Response:
This disclosure adjustment did not have any impact on our nonaccrual policy. The accrual of interest is discontinued and reversed once a retail installment contract becomes more than 60 days past due, and is resumed and reinstated if a delinquent account subsequently becomes 60 days or less past due.
• | Tell us whether your nonaccrual policy changed from prior periods, describe the driver for the change, and tell us how you considered ASC 250 or other relevant accounting literature in making your determinations. |
Ms. Stephanie L. Sullivan
October 27, 2017
Page 16
Response:
This disclosure adjustment did not have any impact on our existing nonaccrual policy, delinquency ratios or credit allowance methodology. We reported delinquencies in the Form 10-Q for the three months ended March 31, 2017 and six months ended June 30, 2017, on a retrospective method to provide investors with comparable data.
• | We note your definition of the “delinquency ratio” on page 62, which is defined as the ratio of end of period delinquent principal over 60 days to end of period gross balance of the respective portfolio, excluding capital leases. Please tell us whether any changes are planned for the calculation of this ratio, in light of the disclosure change to your delinquency buckets (to now be over 59 days past due). |
Response:
As reported in our Selected Financial Data disclosures (Item 2 of our 10-Q filing), we continue to disclose and calculate “delinquency ratio” based on 60 days past due. We have no plans to change the calculation of this ratio, since our nonaccrual policy has not changed. However, we intend to enhance disclosure in future filings, including the “Delinquencies” section of the “Credit Loss Allowance and Credit Quality” footnote, to clarify that delinquency ratio is based on 60 days past due.
12. | We note your disclosure on page 23 that troubled debt restructurings (TDRs) placed on nonaccrual status are considered for return to accrual when a “sustained period” of repayment performance has been achieved. On page 108 of your 2016 10-K filing, your disclosure states that they are returned to accrual “when the account is 60 days or less past due”. Please tell us how you define “sustained period.” If the definition differs from your previously disclosed “60 days or less past due,” please explain the driver for the change and how you considered ASC 250 when making the change and deciding on appropriate disclosures to reflect it. |
Response:
Our accounting policy for classifying loans as accrual, is based on our assessment of collectability of repayment of the principal and interest balances associated with the TDR portfolio. In addition, TDRs are placed on nonaccrual status when the Company believes repayment under the revised terms is not reasonably assured and, at the latest when the account becomes past due more than 60 days.
As noted in question 10, during 2016 we began to note some deterioration in the performance of recent originations, particularly those loans originated in 2015, and addressed those trends with the introduction of more disciplined underwriting standards in late 2016. In addition, more specifically, we began to note deterioration in the performance of our TDR portfolio. We noted in 2016 that while the repayment of the principal and interest related to TDR’s continued to meet our reasonably assured criteria15, there was a continuing negative trend (defaults within the first 6 months of TDR event). Nonetheless we continued to monitor into fiscal year 2017.
15The definition of “reasonably assured” is a matter of judgment. The Company determined to use a threshold of greater than 75% as a measure to determine reasonable assurance.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 17
In 2017, with more data, our analysis confirmed that there had been sustained deterioration in the performance of certain 2015 vintages; that a certain number of deferrals over the life of a loan had a default rate that was continuing to increase to the point of being probable; among other patterns that required us to assess the collectability of interest. The combination of these factors and related data constituted a negative trend and upon reassessment we determined that collection of principal and interest was no longer reasonably assured at the date of the TDR, in certain circumstances. We also anticipated a faster loan delinquency by virtue of the partial payment servicing practice specific to new customers.
Accordingly, effective Q1’2017, we elected to:
• | Retain a loan as a non-accrual loan if at the time of the TDR event it was already on non-accrual status by virtue of being greater than 60 days past due, and further to apply any cash receipts on a cost recovery basis; |
• | Place accrual loans that are designated as TDR in a non-accrual status if they subsequently have a third deferral; and |
• | Not return the TDR to accrual status without an appropriate period of performance (typically defined as six months for a monthly amortizing loan). |
Since some of these TDR loans (as above) were placed on non-accrual, we enhanced our disclosures for return to accrual, to clarify our estimation process to be more closely aligned to (a) regulatory guidance about TDR interest recognition; and (b) disclosures made by SHUSA in its SEC filings.
We concluded that this clarification of our policy, is a “change in estimate” in accordance with ASC 250, and should be applied prospectively only for new TDR’s, based on the following:
• | Our change is driven by a change in the underlying facts and circumstances, notably the collectability of recent TDR vintages and the anticipated impact of servicing practices, which was confirmed in 2017. |
• | With regards to seasoned TDRs, we referred to our earlier analysis and concluded that the historical accounting remained appropriate, since the repayment performance of that loan portfolio supported that collection of principal and interest remained reasonably assured at the date of the TDR. |
Item 2 - Management’s Discussion and Analysis of Financial Condition and Results of Operations, page 51
Provision for Credit Losses, page 66
13. | We note your disclosure that during the three months ended June 30, 2107, you changed the model used for estimating the allowance for credit losses on individually acquired retail installment contracts from the vintage loss model to a transition based Markov model. Your disclosure states that the new model provides data on a more granular and disaggregated/segment basis as it utilizes recently observed loan transition rates from various loan statuses to forecast future losses. Lastly, you state that the change did not have a significant impact on the amount of the allowance for loan losses recognized during the six month period ended June 30, 2017. Please respond to the following: |
Ms. Stephanie L. Sullivan
October 27, 2017
Page 18
• | Describe for us the data that is now provided on a more granular and disaggregated/segment basis, relative to the prior vintage loss model. |
Response
The Markov model provides forecasted gross loss information for TDR and non-TDR separately, on a more granular and disaggregated level as compared to the vintage loss model. Transition matrices in the model are categorized based on account characteristics, such as delinquency status, TDR type (deferment, modification, etc.), internal credit risk, origination channel, months on book, thin/thick file and time since TDR event. Forecasted gross losses for the non-TDR loans are disaggregated into 17 categories based on these characteristics, as compared to the 9 LFS16 bands in the vintage loss model. In addition, the forecasted gross losses for TDR loans are disaggregated into 19 categories in the Markov model as compared to limited categories in the previous vintage loss model.
• | Tell us the potential risks that you now believe this level of disaggregation is able to provide relative to the information based on your prior vintage loss model. |
Response
The Company determined the following benefits of the Markov model in comparison with the legacy vintage loss model:
• | Statistical soundness: The Markov Model provides a more robust and statistically sound framework compared to the legacy models. It recalculates the transition rates every time the model is run, so it is more dynamic in nature and better captures recent fluctuations in the portfolio. |
• | Back Testing: The Company noted that the losses forecasted by the Markov model for both non-TDR and TDR portfolio were much precise, as compared to the legacy models. |
• | More reliable forecasts: The Markov model can account for all probabilities within delinquency and reflects the most recent state of the portfolio as opposed to the legacy models, which rely heavily on origination data. |
• | Detail and flexibility: The Markov model provides granularity and flexibility in handing different kinds of transitions from period to period including transition from non-TDR states to TDR states. |
For these reasons, we decided to adopt the Markov model in Q2’17 to forecast credit losses for use in our credit loss allowance estimate.
• | You state that the change was made during the three months ended June 30, 2017, but state that the change did not have a significant impact on the amount of the allowance recognized during the six month period ended June 30, 2017. Please quantify for us the amount of the allowance change as of June 30, 2017 due to this change in modeling. |
16LFS is an SC proprietary calculation utilized by Risk to assess the credit quality of a particular loan. It is driven by relevant credit metrics (i.e. FICO, LTV, etc.). Lower LFS scores generally represent lower credit quality loans.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 19
Response
The net difference in total allowance for credit losses and impairment for our TDR and non-TDR individually acquired retail installment contacts, as of March 31, 2017 and June 30, 2017, due to change in model methodology (legacy roll rate model to Markov transition matrices) was immaterial, amounting to only 1.6% of the outstanding allowance. Concurrent with the quantitative assessment, we were able to further refine our qualitative factors to align with the specificity provided by the Markov calculation, and the resulting impact was not significant to the overall allowance for credit losses.
• | Tell us whether you evaluated the impact of the use of this model, relative to the vintage loss model, for periods prior to June 30, 2017. If so, describe for us the impacts. |
Response
We evaluated the impact of the change to the new model in Q1’17, as noted above and performed parallel runs for both models for Q1’17 and Q2’17. We analyzed the results from both models comprehensively before switching to the new model in Q2’17. In addition, the results from both models were presented to various committees, in accordance with our governance policies.
• | Tell us whether the new granular and disaggregated/segment basis used for this modeling will also be used for your troubled debt restructuring (TDR) allowance methodology. |
Response
Yes, the new granular and disaggregated/segment (categories) basis used for this modeling is used for both our Non-TDR and TDR portfolios.
Deferrals and Troubled Debt Restructurings, page 123
14. | We note your disclosure that you have $2.6 billion and $1.4 billion of modified loans which are not considered troubled debt restructuring (TDRs) as of June 30, 2017 and December 31, 2016 because they are comprised of loans modified by the Company to adjust the interest rate quoted in a dealer-arranged financing. Your disclosure goes on to state that beginning in the third quarter of 2016, in conjunction with consumer practices, you reassess the contracted APR when changes in the deal structure are made (e.g. higher down payment, lower vehicle price, etc.), and if any changes result in a lower APR, the contracted rate is reduced. Please respond to the following: |
• | In light of the disclosed nature of the deal structure changes, tell us why deal structure changes are occurring so frequently on your retail installment contracts. |
Response:
We implemented a process in Q3’16 of reviewing the final annual percentage rate of charge (“APR”) produced in our funding system versus the actual contracted APR. Before a loan funds, the final contracted structure is run through our automated underwriting to produce a price (APR/discount). At approval the price is based on the specific terms that we approved. The pricing logic is the same at approval and funding, but the dealer may adjust the deal structure between approval and contract.
Due to the granularity of our pricing structure, relatively minor changes in PTI (payment in
Ms. Stephanie L. Sullivan
October 27, 2017
Page 20
income) may result in a pricing change for the deal. When this determination is performed, the customer is current and has not shown signs of credit deterioration. For example, if the structure of the deal has changed (higher down payment, lower vehicle price, etc.) versus the approved APR, then the funding system’s recommended APR is used. Although the customer has agreed to the contracted rate, we reduce the customers contracted rate based on the final calculated APR in the funding system. The full improvement in the rate is passed to the customer. We adjust the payment based on the final rate and notify the customer in writing of the new payment and APR. In the event the final APR produced in the funding system is higher than the contracted rate, this is not passed on to the customer.
• | Please tell us who initiates the deal structure changes. |
Response:
We initiate the deal structure change upon review of final loan documentation.
• | Tell us how long after the original retail installment contract is signed can deal structure changes be made. |
Response:
All deal structure changes occur within 7 days of the date the contract is signed.
• | Quantify for us the impact to the APR rate that you have typically experienced as a result of these deal structure changes. |
Response:
The impact to the APR rate that we have typically experienced over the past 4 quarters ended September 30, 2016 through June 30, 2017 as a result of these deal structure changes is an average interest rate reduction in the range of 51 to 57 bps.
• | Tell us more about the consumer practices during third quarter of 2016 that drove these changes. |
Response:
There was no specific change in customer practices during Q3’16 that drove these deal structure changes. Rather the change was made primarily to give the consumer the benefit of a lower rate due to an improved contracted deal structure compared to the deal structure that was approved during the underwriting process.
* * *
We remind you that the company and its management are responsible for the accuracy and adequacy of their disclosures, notwithstanding any review, comments, action or absence of action by the staff.
Ms. Stephanie L. Sullivan
October 27, 2017
Page 21
The Company hopes that the foregoing has been responsive to the Staff’s comments. If you have any questions related to this letter, please contact the undersigned at (214) 540-2072 or by email at gmartindell@santanderconsumerusa.com.
Sincerely,
/s/ Grace Martindell
Name: Grace Martindell
Chief Accounting Officer
Santander Consumer USA Holdings Inc.
cc: Juan Carlos Alvarez de Soto
Chief Financial Officer
Santander Consumer USA Holdings Inc.
Scott Powell
Chief Executive Officer
Santander Consumer USA Holdings Inc.